Perplexity Blocks Time's AI-Agent Ads, Calling Them Deceptive

A New Ad Format Meets a Hard Block

On August 11, 2026, Digiday reported that Perplexity had blocked every “markdown ad” running on Time.com — sponsored copy embedded in the lightweight, plain-text version of web pages that AI crawlers and agents read instead of the full HTML site humans see. The answer-engine company did not treat this as a routine ad-policy dispute. It called the practice deceptive, full stop.

This is, as far as trade coverage shows, the first public fight over advertising aimed at AI agents rather than people — and it lands directly on a fault line that search engines have policed for two decades: cloaking, the practice of serving different content to crawlers than to human visitors.

The move matters because it draws the first real battle line in what could become a major new advertising economy. If agent-facing inventory is legitimate, publishers watching bot traffic overtake human traffic have a whole new revenue line to build. If it is cloaking by another name, early movers are gambling their standing in the AI search indexes that increasingly decide who gets cited at all.

How Markdown Ads Work

Time’s program, launched in late July 2026 with ad-tech firm Mobian, is mechanically simple. Mobian takes a brand brief and writes FAQ-style sponsored copy — question-and-answer formatted text designed to read like editorial content. That copy gets seeded into the markdown rendition of Time’s pages. AI crawlers ingest it; human readers never see it, because browsers render the full HTML version.

Mobian then tracks which AI search engines pick up the copy and how warmly they treat it. Ally Bank and the Project Management Institute were the first advertisers to buy in.

Time marks the placements as sponsored at the top of the page — a disclosure that exceeds what any current rule requires. But once an agent ingests the raw markdown text, that label falls away. The machine reading the content has no reliable way to know a human-facing disclosure ever existed.

That detail is the crux. Perplexity’s objection is not to sponsorship or labeling per se. Perplexity’s chief communications officer Jesse Dwyer told Digiday the company works “continuously” to protect users from “deceptive practices, sponsored or not.” A sponsor label, in Perplexity’s stated view, does not rescue a placement that humans cannot see at all.

The Traffic Logic Behind the Format

Publishers are not inventing demand for agent-facing formats out of thin air — the traffic data is forcing their hand. Digiday reports that Time now sees more bot than human traffic on most days. Cloudflare’s measurements put bots above 50% of all web traffic globally.

When the majority of “readers” fetching your pages are machines, the classic display-ad model — impressions served to human eyeballs — structurally leaks value. Every crawl that ingests your content without rendering a single banner is, from a publisher’s perspective, a free ride for the AI ecosystem. Markdown ads are an attempt to monetize exactly that crawl traffic.

The commercial stakes scale with query behavior, too. The Digiday reporting notes that agent-driven activity is growing as a share of how people find information — which means the citation layer that Perplexity, ChatGPT Search, and similar engines control is becoming the new front page. Being blocked there is not a rounding error; it is a threat to discovery itself.

Cloaking: The Precedent That Matters

Serving crawlers different content than humans is the textbook definition of SEO cloaking — a violation that Google spent years punishing with manual penalties and delistings. One consultant quoted in the coverage raised the parallel directly, and it frames the dispute better than any “ad format innovation” framing.

The classic cloaking scam served keyword-stuffed junk to Googlebot and a normal page to users. Time’s markdown ads invert the incentive — sponsored copy goes to the crawler, clean editorial goes to the human — but the structural shape is identical: two audiences, two versions, one of them secretly monetized.

Perplexity’s enforcement lever is its trust score, a proprietary ranking signal in its search index. Dwyer warned that publishers deploying “deceptive advertising like markdown ads” risk a “reputational downgrade.” Digiday also reported that Perplexity’s search and security teams are working on additional, broader measures against markdown ads — suggesting the block on Time.com is a first instance, not a final policy.

Notably, Perplexity would not explain to Digiday what specifically counts as deceptive, or how the blocking technically works. The line is drawn, but the ruler is private.

The Irony Problem

Perplexity’s moral authority here is contested. Critics were quick to note the irony of a company built on crawling publishers’ content — often without licensing deals — now lecturing publishers about deceptive practices. Shopifreaks’ coverage of the block openly enjoys “the irony of Perplexity calling out anyone else for their ‘deceptive’ practices.”

The contradiction is real but cuts both ways. Perplexity has faced publisher lawsuits and criticism over how its crawlers summarize content it did not license. At the same time, its interest in a clean citation layer is structurally aligned with users: if the source layer an answer engine relies on can be secretly bought, every answer it produces is contaminated. An answer engine that tolerates agent-ads is an answer engine whose citations are for sale.

There is also a deeper business logic. Time’s COO Mark Howard, discussing the program’s launch in July, acknowledged the uncertainty: “We don’t know yet because this is b…” — his on-record quote trails off into exactly the kind of experimentation this dispute now forces to a resolution. Perplexity blocking the format early, before a dozen more publishers adopt it, is cheaper than unwinding it later.

A Risk Ladder for Agent-Facing Tactics

The dispute suggests a rough spectrum of agent-facing tactics, from safest to riskiest:

1. Labeled, human-visible sponsorship — ads that appear in both the human and crawler renditions, clearly marked. No cloaking; standard disclosure norms apply.
2. Labeled, crawler-only copy — Time’s current format. Disclosed at the top of the markdown, but the disclosure is invisible in the human render and degrades on ingestion.
3. Unlabeled, crawler-only copy — outright astroturfing of the citation layer. Universally treated as spam.

Perplexity’s “sponsored or not” phrasing implies it treats category 2 and 3 the same — the dividing line is whether humans can see what agents are being fed, not whether a label exists somewhere in the raw text. That is a stricter standard than most publishers anticipated.

What Happens Next

Three open questions will shape the outcome. First, whether Google and OpenAI follow suit — ChatGPT Search and Google’s AI Overviews face identical incentives, and a coordinated de facto standard would kill the format outright. Second, whether Time (which did not respond to Digiday’s request for comment on the block) fights back publicly, adapts the format, or quietly drops it. Third, whether any regulator weighs in: the FTC has shown appetite for policing undisclosed sponsorship, and “ads only machines can see” is an unusual posture for existing disclosure rules.

For publishers, the near-term lesson is conservative. The traffic shift toward agents is real, and monetizing it is a legitimate problem. But building that monetization on a two-version page — one for humans, one for machines — recreates the exact structure that search engines have treated as spam for twenty years. The formats that survive will be the ones where the human and the agent see the same thing.

For everyone else, the episode is an early glimpse of the governance layer forming around AI-mediated information: private companies, wielding proprietary trust scores, deciding which persuasion attempts reach the machines that answer your questions. That power is now visibly in use — and its first target was one of the most recognizable media brands in the world.

One caveat worth keeping: virtually all original reporting here traces to a single Digiday article. Perplexity published no policy post of its own, Time has not commented on the block, and secondary coverage re-quotes the same sources. Treat details as accurately reported but single-sourced.